Digital + Analog: A Shotgun Wedding
originally published in L’Amour Report of November 2003
It is still possible, although difficult, to build a totally “analog” studio. Analog technology can do the job without relying on ANYTHING digital. You CANNOT build a totally digital studio. Don't believe me? Just try it.
OK, how about a digital recorder? Even that is not totally digital. For example, the ADAT recorder has A/D converters. Even the name “Analog to Digital converter” should be obvious, since it has to START with an analog section. The same “problem” applies to stand-alone Hard Disk recorders. Without A/D converters, you cannot record any analog source. Need a microphone for vocals? Forget it. Forget recording any acoustic instrument.
The only source you can record digitally without converters is some kind of synth or sampler with digital outputs. Oops... did I say sampler? How did you get those samples in the first place? Somebody had to set up a microphone.... Oh. Right. You CAN record a digital synth that does not use samples. You can, in fact, make lots of music that way. You can sequence and record as many tracks as you want.
Now I'm going to go Zen on you: If a tree falls in the woods, and there is no one there to hear it, did it make any noise? If you create, record, and mix music, but you don't hear it, can you be sure you got the result you want? Sure, you could argue that Beethoven composed some of his greatest work after he was pretty much stone deaf, but who wants to be deaf? Without a D/A converter, not to mention headphones or speakers, you will hear NONE of your music. In order to be useful for making music (or any other kind of sound recording), digital technology must be “married” to analog technology. This wedding is not optional, it is necessary; You might call it a shotgun wedding.
At the very least, the necessary analog stuff in your studio includes converters and monitor speakers. Monitoring entirely with headphones does not adequately tell you what you (or someone else) would hear through speakers, of course. There have been a few active monitor speakers made with their own built-in A/D converters, which can be convenient in some situations. Generally, though, I would rather have the speakers and converters be individual pieces of hardware. For one thing, you may want to upgrade either the speakers or the converters at some point without having to replace both at once. Besides, if you keep more than one set of speakers in your studio, isn't it silly to have another set of converters for each pair of speakers?
I also want the monitor volume control to be in the analog part of the chain (even in most digital mixers the monitor volume control is AFTER the converters). If the volume control is in the digital part of the chain, you have to sacrifice resolution to listen at lower volume; For every 6 dB drop in monitor level, a digital volume control has to give up one bit of resolution. Many of us routinely listen at least 30 dB below the maximum possible volume, which, if we must digitally control volume, will cost us 5 bits of resolution. Dropping the volume of a 16 bit recording (digitally) by 30 dB would then be like listening to an 11 bit recording.
The recommended monitor chain, then, would at least include the A/D converter, some kind of monitor control unit (I have used, for example, an Adcom GTP-500 II preamp to let me select monitor sources and control volume), and at least one set of monitor amplifiers and speakers (I don't mind combining the amps and speakers, as active monitors can have some real advantages). The goal of the monitoring system is, of course, the same as it was with the old analog studios: letting you hear the sound you are creating without contributing a sound of its own. The single word that describes this ideal is “transparency”. Think of a window. If it is made with clear glass, you can see clearly what is going on outside. A stained glass window, on the other hand, exists to show you its own picture, rather than a view of what is on the other side. Since there is no such thing as a perfectly transparent speaker we do the best we can, sometimes even using different sets of speakers to test how our mixes might sound in different places.
The concept of transparency applies to the entire recording chain, of course. Some of the designers of microphones and preamps, in particular, continue to strive for products whose sound is as “uncolored” as possible. With our analog gear, though, we have come to look for both “kinds of windows”. While most of us try for transparent monitoring systems, at least, and some of us look for transparency in our mics and preamps as well, many of us also look for analog equipment that we think of as being in the “stained glass” category. We buy things that will let us add more pretty colors to the sonic images we create. This is why microphones, preamps, equalizers, compressors, and even tape recorders using vacuum tubes have become popular again. Not too many years ago many audio professionals looked upon any old gear that used tubes as basically junk. For a while, things like the Teletronix LA-2A compressor and the Pultec EQP-1A equalizer could be had for less than a hundred dollars: now prices for these old units are up in the thousands.
Whether we want our analog windows to be clear or colorful, a lot of the quality of the sound we finally get depends upon how all these things are put together. Even the way that the analog and digital parts are tied to each other is important. If the wedding is peformed well, the marriage will not only last but the desired fidelity will be there too. This marriage that we are talking about is usually called the “interface” between different pieces of equipment.
On The Level
Sometimes in equipment manuals you will see the inputs or outputs being described as a “pro” or “consumer” interface, with a specified “reference” level. The most common reference levels used are “+4dBu” for “pro” equipment, “-10 dBv” for “consumer” equipment, and “-20 dBv” (sometimes called “instrument level”, although some musical instruments put out signals that are MUCH hotter). A/D and D/A converters are definitely “interfaces”, and their analog inputs/outputs are also designed to work at specific reference levels.
This idea of reference levels can be tricky. Apparently it originated with telephone engineers a long time ago. They were designing systems to carry speech long distances, so they needed a way to look at speech signals with a meter. The meter they developed became the standard Volume Unit meter, and the “0 VU” reference level was marked at point about two thirds of the way up the scale. This reference level basically was the best level for a signal, “hot” (high) enough to be well above the noise, but not so hot that it would overload the circuit. When tape recorders came into use, this reference level idea was applied in much the same way as for phone circuits, as a way to see that you were recording far enough above the noise without overloading the tape.
Digital recording levels are, in a way, simpler than what we do with analog. Analog tape recordings gradually get more distorted as the level goes up (more about this later). With digital, recordings get more distorted as the level goes DOWN. When the recording resolution is 16 bits or more, you don't always have to “crowd” the level to the very top, but higher levels are still “cleaner”. Of course, once you hit the top, that's all there is. There is nothing gradual or graceful about the way digital overloads. The maximum digital recording level is referred to as Zero Decibels below Full Scale, or “0 dBfs”. This is NOT the same as “0 VU”!
Why should 0 dBfs be different from 0 VU? For one thing, in analog recording the best recording level is somewhat below saturation, unlike digital's “as hot as possible” preference. Another important reason is the kind of metering we use. Digital recorders use “bar graph” style meters. These meters have no moving parts, so they can react instantly and accurately to signal peaks. Most professional analog recorders use mechanical VU meters as level indicators. That little word “mechanical” is important, since it means that the meters have physical moving parts.
If you drive a car, you know that anything heavy takes time to get moving, and it takes time to stop. The needle pointer of a VU meter may weigh a lot less than a car, but it still needs time to get moving. If you suddenly hit a VU meter with a “0” level tone, it will take the needle almost a third of a second to reach 0 VU. Any level that is not held for at least that long will not be fully shown by the VU meter. A snare drum hit, for example, is very fast, fast enough that by the time the meter “peaks” the signal is already gone, and the needle can't get high enough, so the level is actually higher than what the meter says. The engineers who designed the meter in the first place knew this, but back then a meter that could read peaks accurately could not be made easily portable.
Most speech signals, fortunately, last longer than a snare hit, so the engineers concentrated on a meter design that was “good enough”, at least fast enough, for speech. Even then the meter could not show all of the highest (short) peaks, so the engineers deliberately left some “headroom” for signals that peaked as much as 10 dB above 0 VU. Many recording consoles have at least 20 dB of headroom above 0 VU.
Many A/D and D/A converters are designed for a “+ 4 (dBu)” reference (0 VU reading) level. The headroom of these converters may be anything from 12 dB to 20 dB. The converters I use, for example, have 15 dB of headroom, which means that 0 VU = +4 dBu = -15dBfs. Looked at another way, for my converters +19 dBu = 0 dBfs.
So why is this particular bit of Tech Talk Trash important? Part of making the equipment interfaces work is making sure the levels “match”. More exactly, we want to get the most dynamic range from our system that we can. Dynamic range involves two questions: First, how soft can the sound get before it gets below the “noise floor” (hiss level)? and second, how loud can the sound get before it hits “crunch” level (0 dBfs in our digital stuff)? The difference between these two levels is the dynamic range (in dB) of the system. Each box in our system has its own dynamic range, set by its noise floor and its clipping (distortion) level.
Imagine that the dynamic range of each piece of equipment is a window. The bottom of the frame is the noise floor, and the top of the frame is the clipping level. Now let's suppose that the window of “Box A” is 90 dB high, and the window of “Box B” is also 90 dB high. If we connect the two together, the dynamic range of our “system” will be 90 dB, right? Not necessarily. This is where level matching, often called “gain structure” comes into play. If Box A has a clipping level of +10 dBu, and Box B clips at +20 dBu, the “windows” may be the same size, but they don't match up. The top of Window B is 10 dB higher than the top of Window A, and the bottom of Window B is also 10 dB higher than the bottom of Window A. Each window may be 90 dB high, but when the windows are sandwiched together the total opening (where the individual openings overlap) is only 80 dB.
Quite often the dynamic range windows of different pieces of equipment are of different sizes. When you chain two or more boxes together, the best possible dynamic range for the whole system is no bigger than the smallest window. If Box A has a window only 70 dB high and Box B has a window 90 dB high, the maximum dynamic range for both boxes together will be limited by Box A, which has the smaller window. In this case, Box B having a 10 dB higher clipping level than Box A does not make the available dynamic range smaller, since having the top of Box A's 70 dB window 10 dB below the top of Box B's 90 dB window still leaves the bottom (noise floor) of Box A's window 10 dB higher than the bottom of Box B's window. As long as both ends of Box A's window are “inside” the ends of Box B's window, we still have the best dynamic range we are going to get.
In our example, Box A could be a microphone preamplifier and Box B could be an A/D converter. Many A/D converters have dynamic ranges of 90 dB or more (the best converters with 20 bits or more of resolution can do somewhat better than 100 dB). Some (especially older) analog gear can have a dynamic range of 70 dB or less.
As a general rule, when the dynamic range windows are different sizes, it is best to match the TOP edges of the windows, especially if the bigger window belongs to the A/D converter. Why? First, because the resolution of the converter is best at higher signal levels. Second, because when mixing digitally (as many, if not most, of us do now) it is often easier to turn tracks down than it is to boost them. Of course, by bringing up the input level you are also pushing up the noise. Often this will not be heard in the mix. If the noise IS a problem, you may want to process it out using a plugin or some program to process the file. If you use noise reduction of this kind, it can do its job better when it has more bits to work with, so feeding it a higher level can help.
I need to mention one possible problem that some converters have that can limit how hard you want to “push” them: sometimes certain signals that go above -6dBfs can be distorted. The cause of this problem is the filtering that the converter has to use to keep it from generating “notes” of its own that you do NOT want. These filters have to be very “sharp”, meaning that they cut off above a certain frequency very quickly. Filters like this can “ring” when they are hit hard enough, especially at high frequencies (often produced by cymbals and other percussion instruments). This ringing will often peak ABOVE the level of the signal, and if the signal is hot enough the ringing can be clipped even when the signal is not. This problem is more likely to happen with the cheaper converters, but some of the more expensive converters may also have it. If you are not sure, you may want to take extra care with cymbals, percussion, and other high-frequency sounds.
So how do you make sure that levels are properly matched in your system? It is best to start with your A/D converters. Some converters have input level adjustments, but many do not, so you usually have to set up your other gear to match the converters. Often the user manuals that come with the converters will tell you what the maximum input level is. In my system, for example, that level is +19 dBu. There is almost always some kind of metering available on either the converters or the recorder (stand-alone like an ADAT or DAW as in ProTools) that will tell you what the level is going into the converters. Any piece of analog gear that can be “turned up” past the point where the digital signal meters go all the way to the top is able to overload the converters. If so, you just use the digital meters to tell you when you need to back off on the level.
If the digital meter stops going up before the volume on the analog box is all the way up, there is a good chance that the analog device cannot drive the converters quite hard enough, and a real risk that you will have clipping in the analog part of the chain when you are trying to get a decently high recording level. This is most likely to happen if the converters are designed for “pro” (+4 dBu reference) levels and you are feeding them with stuff that puts out “consumer” (-10 dBv reference) or “instrument” (-20 dBv reference) levels.
One example would be when you are trying to sample sounds from some vinyl. Your turntable is probably hooked up to a home stereo setup, and the easiest way to feed it into your digital rig is to take the signal from the tape recording outputs of your stereo. Unfortunately, the tape outputs generally assume a reference (0 VU) level of -10 dBv, some 12 dB lower than what pro gear expects to get, and the preamp driving them may very well clip well before your digital input meters read 0dBfs. More likely, since the gain from the phono input to tape output is not adjustable, the recording level will just be lower than you want. The simplest (and probably best) solution here is to feed the tape outs through some other box, like a small mixer or a compressor (you usually CAN set up a compressor to give you gain without any compression), that has “pro” level outputs.
The Buzz Monster
So far, when I say “noise floor” I mean “hiss level”, but hiss is not the only noise problem you can have. Back in the day, when you looked at the specs for stereo gear, there was a specification called “Hum and Noise”. These two things were lumped together because the meters they used to check it could not separate the hum from the hiss. Most of the equipment we now use has a hum level well below the hiss level, unless you hook it up the wrong way. One of the most important ways that we keep hums and buzzes out of our stuff is to use equipment that has “balanced” inputs. These inputs “balance out” (self-cancel) stray hums and buzzes from grounds and wiring IF they are wired correctly.
The active monitor speakers I use in my studio have balanced inputs. The preamp that drives them has UNbalanced outputs. When I first got the speakers, I tried using regular “guitar” cables to feed them from the preamp. The speakers sang. Well, actually, they didn't know the words, so they hummed. One note, and I bet you know which. I had to make special cables, basically by hacking up a couple of microphone cables. My speakers have XLR inputs in the back, which would take the male ends of the mic cables. I cut off the female ends of the mic cables and put regular guitar plugs on them, with the low (formerly pin 3) side conductors tied to the sleeves of the plugs along with the shields. Result? No more hum. I had to do a similar thing feeding the unbalanced outputs of some synths into the balanced line inputs of my board, only this time instead of a male XLR connector I had a 1/4” TRS connector to the line in of the board. Again, the hum went away.
Moral? Always use balanced wiring to feed your balanced inputs, even if it is coming from unbalanced outputs. Of course, balanced out to balanced in is always best if you can get it.
The Pretty Colors
So far, I have focused on the “transparent” uses of analog gear. Now we can pull out the “stained glass”. This is when we want the equipment to color or change the sound.
Some equipment was designed deliberately to change the sound. The more obvious examples would include reverbs, delays, and all sorts of “processors”, including the “stomp boxes” (effect pedals) used by guitar players. Generally the rules for tying these processors into our studio system are the same as what we do for the “transparent” stuff. We try to feed the box the level it wants to see, and if the level it puts out is not right for the next box, we boost or drop it as needed.
Most guitar stomp boxes, for example, “like” a signal to be at about the -20 dBv level, both going in and out. Some of these can accept higher levels just by turning down a level control, which would make input interfacing fairly easy. Sometimes they can also put out higher levels, but commonly you will need to boost the output by going through something like a small mixer or some other device.
These boxes usually have unbalanced inputs and outputs, so you may need a signal isolation transformer. Radio Shack sells a fairly cheap “audio system ground loop isolator” (Cat. No. 270-054) that may do the trick. The Furman IP-2 “Isopatch” is a somewhat more expensive unit that may also work well. There are a number of others, and I only mention these two because I found them lying around my shop (I don't own any stomp boxes, so I haven't tried these transformers that way).
The other equipment we might use for “color” was not intended to be used that way. That old stuff with tubes in it that we like because it sounds “warm” was pretty much the only game in town when it was made. Until at least the late 1960's almost all recording studios used tube-type equipment exclusively. The goal of the people who designed and built this stuff was to capture sound as cleanly, accurately and transparently as possible. Many of the things we now like about “tube sound” were actually defects that the design engineers did their best to eliminate. Some of this old equipment, used the way it was intended, is still amazingly accurate.
Some of what we do to get neat sounds out of this old gear amounts to abuse. The early guitar amplifiers, for example, were actually designed to be fairly clean but cheap. The cheapness is part of where the distortion came from, but the way we often use these great old amps now is to drive them much harder than the amp builders intended. The tube guitar amps built now, of course, are designed to make distortion easy to get, but this happened only after so many guitar players abused their amps that the manufacturers turned a defect into a feature in self-defense.
It isn't much of a surprise, then, that we get some of the coloration we want by abusing old (or even new) tube equipment. We studio guys do pretty much what the guitar guys have done for decades: push the equipment into distortion. One engineer, for example, reports that his favorite way of getting a fat direct bass sound is to run the bass DI into a mic preamp, and the mic preamp into an old LA-2A limiter. He cranks the output level on the LA-2A much higher than normal, then drops the level with a “pad” (attenuator, which is kind of like a volume control turned part way down, only there is no knob to turn) between the LA-2A and his A/D converter.
A similar trick can be done with the output of most studio tube gear. The first part, of course, is that we are pushing a much higher level so that the box starts to distort. As a result, the output level of the box is much higher than normal. A lot of old studio tube gear is designed so that its output level at “clipping” (where the fuzz kicks in) is about +30 dBu. This is easily at least 10 dB higher than the overload level of many A/D converters, so we need to cut it down a bit. Turning down the level inside your tube limiter or whatever would work, but then you wouldn't be driving the output amp the way you want to, so we need to drop, or attenuate, the level after the output. This is what a pad (attenuator) does. The pad is basically 3 resistors wired together in a “U” shape, with the signal going into the “top ends” and the output taken from the bottom corners. If you are good with a soldering iron, it is easy to make. The output from the pad, of course, then goes to the A/D converter.
There is another neat side benefit of “padding” the output: Dropping the signal also drops the noise. In other words, by padding down the signal to prevent overload, we also got back some lost dynamic range. Basically, we just pulled down the “window” of Box A to that its top is even with the top of the window of Box B.
There is one more class of “colored glass” to cover here: Analog tape. A fair number of studios now use analog tape recorders as “processors” to color the sound. This is another case of the imperfections becoming the virtues we want. Like some tube equipment, analog tape distorts in a gradual way, getting “worse” as the level goes up until it “saturates”, which is the analog tape version of clipping, only a bit softer. There are actually a few different things that analog tape does to sound, but I won't try to list or explain them all here (that would be an article by itself). The main point here is that analog tape has a sound that is not exactly like anything else. A few people have developed digital effect plugins that are supposed to duplicate the sound, but as far as I know no one has really modeled all of what analog tape does.
Connecting an analog tape recorder into a “digital” studio is fairly simple, since its inputs and outputs are electrically much like most other equipment. All of the same methods apply as with other gear. The main difference is that there is a time delay to deal with that complicates things a bit. First of all, there is more than one way that you can use the machine. The obvious thing to do is record the tape, then rewind it and play it back into your digital recorder. It is also possible, though, with proper professional tape machines, to do both processes at once. The tape machine has three “heads” that get the sound onto and off of the tape. When the tape is running, it first passes the “erase” head, which removes any earlier recording from the tape. Next it passes the “record” head, which “writes” the new audio to the tape. Finally it passes the “play” or “reproduce” head, which reads the recording from the tape. The use of three heads makes it possible to listen to the tape as it is being recorded. This was first done so that a recording enginer could listen to the tape as it was being recorded to make sure that the recording was OK. If there was a defect in the recording, the engineer would hear it right away and could stop the session for another try.
We can still run the machine this way, but this time so that we can feed it into the digital recorder. This way the process is done all in one pass of the tape, instead of running the tape twice. Let's say there is one track you want to process. Here's how:
1) Assign the track to its own individual output.
2) Connect that output to the tape recorder input.
3) Connect the playback output of the tape recorder to an input of the digital recorder.
4) Send that input to its own fresh track.
5) Set the tape recorder monitor mode to “repro”.
6) Put the tape recorder into “record ready” mode.
7) Start the tape and punch into “record”.
8) Start the digital recorder, playing the source track and recording the new track.
When this is done, you have a “processed” copy of the original track on the digital recorder. The only remaining problem is that the processed track is now a fraction of a second late. If the digital recorder uses tape, it may have a delay compensation feature that you can use (the user manual should tell you how). If you are using a DAW (such as ProTools), you can “slide” the new track to line it up with the original.
It is also possible to process a live source while tracking, and in some cases you may prefer this. Instead of playing a track from the digital recorder, you feed your source directly into the analog tape machine (instead of the first two steps listed above) and record the output of the tape machine to your digital recorder at the same time. You need to figure out how far to slide the track afterwards, of course, and while recording you need to monitor your input source from BEFORE the analog recorder (otherwise the performer will get VERY confused).
By the way, although any 3 head analog recorder can be used for this trick, you will get the best results with a good professional machine that is in good condition and set up (aligned) properly.
Remember that the analog part of your studio is vital to the quality of your work. Use the best equipment you can afford, and put your system together properly. Even the best equipment, if poorly wired and poorly used, will not give you the best sound that it is capable of. Still, don't be afraid to try new methods, and above all, have fun. Making music is too much work to have to be miserable doing it.